08-17-2017, 12:22 AM
Abstract:
Developed a system for answering drug information related queries using machine learning based classification to classify the query into one of the pre-determined set of classes. The query is passed through various query preprocessing modules like Spelling Correction, Abbreviation Expander and Stemming following which it was passed through a Shallow Parser and enriched with WordNet synonyms and UMLS concepts for feature generation. After classification, answer is retrieved from a comprehensive database. Finally the system was successfully implemented for sending/ receiving natural language queries/ answers via SMS. Worked during Summers 08 on the project under SURA (Summer Undergraduate Research Activity). Implemented in Java.